Split word with condition

I would try REGEXP_REPLACE() as a function:

WITH input(src) AS (
          SELECT 'agricultural & manufacturing reforms'
UNION ALL SELECT 'manufacturing & construction sectors'
)
SELECT
  src
, REGEXP_REPLACE(src,'(\w+) \& (\w+) (\w+)','\1 \3 & \2 \3') AS tgt
FROM input;

src | tgt |
-------------------------------------+---------------------------------------------
agricultural & manufacturing reforms | agricultural reforms & manufacturing reforms
manufacturing & construction sectors | manufacturing sectors & construction sectors

It helped! and to identify such data I used REGEXP_COUNT